WCDMA for UMTS

Luc Vandendorpe

UCL Communications and Remote Sensing Lab.

1

Introduction

• Second generation (like GSM) enabled voice traffic to go wireless

• In several countries there are now more mobile phones than landline(wired) phones

• Data handling capability of 2nd generation is limited

• Third generation: should provide high bit rate services that enabletransmission and reception of high quality images and video and pro-vide acces to the WEB

• Third generation: referred to as UMTS (Universal Mobile Telecom-munications System)

2

Introduction

• WCDMA (Wideband CDMA) is the main third generation air interface

in the world

• Specification created in 3GPP (Third Generation Partnership Project)

which is joint standardisation project of Europe, Japan, Korea, USA

and China.

• In 3GPP WCDMA is called UTRA (Universal Terrestrial Radio Access)

FDD and TDD

3

Introduction

• Work for 3rd generation started in 1992 when the WARC (WorldAdministrative Radio Conference) of the ITU (International Telecom-munication Union) identified frequencies around 2 GHz for 3G

• Withing the ITU, 3G was named IMT2000 (International Mobile Telephony-2000)

• The target of IMT was to have a single worldwide standard

• WCDMA will be used by Europe, Japan, Korea in the WARC-92spectrum allocated for 3G

• In North America, spectrum auctioned for 2G. No specific spectrumfor IMT-2000

4

Introduction

• Other standards can be used for 3G: EDGE (Enhanced Data Rates

for GSM Evolution) and Multicarrier CDMA (CDMA 2000)

• EDGE: up to 500kbps with the GSM carrier spacing of 200kHz

• Multicarrier CDMA: enhancement of IS-95

5

Introduction: situation in the world

6

Spectrum/band allocation

• In Europe and most of Asia, the IMT-2000 bands (2×60GHz) (1920-1980 and 2110-2170 MHz) will be available for WCDMA FDD

• About TDD: in Europe it is expected to be in 1900-1920 and 2020-2025 MHz (25 MHz in total)

• Japan and Asia: FDD bands like in Europe

• In Japan part of the TDD spectrum is used for PHS (cordless tele-phone)

• In the USA no spectrum left for 3G. The existing PCS spectrumwill have to be refarmed to allow 3G; EDGE has an advantage therebecause it can be deployed in GSM 900 and 1800 bands

7

Spectrum/band allocation

8

Differences between WCDMA and 2G

• Bit rates up to 2Mbps

• Variable bit rate to offer BW on demand

• Multiplexing of services with different quality requirements on a singleconnection (speech, video, data)

• Quality requirements from 0.1 FER (frame error rate) to 10−6 BER

• Coexistence of 2G and 3G and inter-systems handovers

• Support of asymmetric uplink and downlink traffic (like ADSL: WEBimplies more downlink traffic)

9

Differences between WCDMA and GSM

10

Differences between WCDMA and GSM

• Larger BW required to support high bit rate

• GSM covers also services and core network aspects. The GSM plat-

form will be used with the WCDMA air interface

• Transmit diversity is included in WCDMA to improve downlink capac-

ity to support asymmetry

11

Differences between WCDMA and IS-95

12

Differences between WCDMA and IS-95

• Larger BW of WCDMA gives more multipath diversity, especially insmall urban cells

• WCDMA has fast closed-loop power control in both UL and DL; IS-95has only in UL. in DL it increases DL capacity

• IS-95 targeted mainly for macro-cells and use GPS sync of the base-stations; more dfficult in non LOS environments

• WCDMA uses asynchronous base-stations; it impacts handover

• Also possibility of having inter-frequency handover in WCDMA (sev-eral carriers per base-station); not specified in IS-95

13

WCDMA description: spreading and modulation

• Spreading is used in combination with scrambling

• Scrambling: used on top of spreading; needed to separate terminals

or base stations from each other

• Scrambling does not change the chip rate nor the bandwidth

14

Channelization codes

• Transmissions from a single source are separated by Channelization

codes: downlink connections in one sector or the dedicated physical

channels in the uplink from one terminal

• Based on OVSF technique (Orthogonal Variable Spreading Factor)

• Allows spreading to be changed while maintaining orthogonality be-

tween codes

15

Channelization codes

• See code tree

16

OVSF

• Generation process inspired from Hadamard codes

• Restriction: when a code is intended to be used, no other code gen-

erated from the intended code can be used (as for higher SF); no

code between the intended coded and the root can be used (as for

smaller SF)

• Restrictions apply to individual sources; do not apply to different base

stations (separation by scrambling) or to different mobiles in the uplink

(separation by scrambling)

• Chip rate: 3.84 Mchips/sec (subject to changes)

17

Channelization and scrambling codes

• Channelization (OVSF)(Increases BW)

– UL: separate DPDCH (Dedicated Physical Data CHannel) from

same terminal; DL: separate connections to different users in one

cell

– UL: 4-256 chips; DL also 512

• Scrambling (Does not hange BW)

– UL: separate terminals; DL: separate sectors

– UL: if long code, 38400 chips (Gold codes); if short code (JD),

256 chips (extended S); DL: 38400 chips

18

Uplink

• For the terminal, maximise terminal amplifier efficiency and minimizeaudible interference due to terminal transmission

• When no speech information, time multiplexing of power control info(1.5kHz) would lead to audible interference

• Therefore the two dedicated channels (data and control) are I-Q codemultiplexed instead of time multiplexed

• Pilot and control are maintained on a separate continuous channel;no pulsed transmission

19

Uplink

• For the best power amplifier efficiency, PAR (peak to average) ratio

should be as low as possible (minimal back-off)

• With I-Q code multiplexing called dual-channel QPSK, levels of DPDCH

and DPCCH (Dedicated Physical Control CHannel) are different

• When data rate increases (to maintain identical Eb) could lead to

BPSK-like transmission (unbalanced)

• Therefore complex spreading is used to ”share” I-Q info with the two

branches

20

Uplink

• Two possible constellations before scrambling (depending on the level

of the pilot)

21

Uplink

• Efficiency remains the same as with balanced QPSK

• Efficiency of the power amplifier does not depend on G

• Power difference between DPDCH and DPCCH quantized to 4 bits

22

Uplink spreading

• When code used by the DPCCH, the same code cannot be used evenof a different I or Q branch

• Would otherwise interfere with the phase estimation achieved withDPCCH

• Spreading factor of the DPDCH can change on a frame basis

• There is a single DPCCH per radio link; but there may be severalDPDCH

• For DPDCH the same OVSF code can be used on different I-Qbranches

23

Uplink spreading

24

Uplink frame structure for DPDCH/DPCCH

• 1 frame=10msec=15 slots=15 × 2560 chips=15 × 2560/(256/2k) =

15 × 10 × 2k bits (SF=2k, k = 0, · · · ,6)

• For DPCCH, SF=256 always hence always 10 bits per DPCCH slot

• TFCI: rate information; TPC: transmission power control; Pilot: for

channel estimation

25

Uplink frame structure for DPDCH/DPCCH

• Advisable to transmit with single DPDCH as long as possible (for PAR

reasons)

• With single DPDCH:

– 960kbps can be obtained with SF=4, no coding

– 400-500 kbps with coding

• With 6 codes, up to 5740 kbps uncoded or 2Mbps (or even more)

with coding

26

Uplink frame structure and bit rates

27

Uplink scrambling codes

• Specific to each source

• Long codes: truncated to the 10 msec frame length; hence 38400

chips (if 3.84Mcps) (code used for real part, and the same for imagi-

nary but with delay)

• Used if base station is rake based

• Short codes: 256 chips (two codes used for real and imaginary parts)

• Used if joint detection or interference cancellation is implemented

28

Downlink modulation

• Normal QPSK with time multiplexed control and data streams

• Audible interference not an issue because common channels have con-

tinuous transmission

29

Downlink spreading

• OVSF based like in the UL

• Same real code for I and Q bits

• Code tree under a single scrambling code shared by several users

• One scrambling code and hence code tree per sector

• Common channels and dedicated channels share the same code tree(one exception for the SCH-synchronization channel)

• Channel spreading factor does not change on a frame-by-frame basis

30

Downlink spreading

• Variable bit rate taken care of by rate matching or discontinuous

transmission

• If multicode transmission for a single user, parallel code channels have

different channelization codes

• SF all the same with multicode transmission

31

Downlink scrambling

• Long codes like in the UL (complex spreading)

• Code period truncated to 10 msec (to ease the task for the terminalto find the right code phase, with a 31 degree code generator)

• Primary set of 512 codes; if needed, secondary set of 15 codes perprimary code, meaning 8192 codes in total (512 × 16)

• Before the terminal synchronizes with the cell spreading code it mustsynchronize with a code word identical for all cells (actually a firstcode common to all cells and a second specific to groups of cells;only needed at power-on)

• No prior timing is available so matched filtering must be implemented

32

Downlink

• Summary

33

Downlink frame structure and bit rates

34

Rate matching

• Used to match the number of bits to be transmitted to the number

available in a single frame

• Achieved by either puncturing or repetition

• In UL repetition is preferred; puncturing only used when terminal

limitations are reached, or to avoid multicode transmission

• May change on a frame-by-frame basis

35

Channel coding

• Two methods:

– 1/2 rate and 1/3 rate convolutional coding for low data rate

services (like in 2G)

– 1/3 rate turbo coding for higher data rate services

∗ 8 state PCCC (parallel concatenated convolutional code)

∗ minimum blocks of 320 bits should be processed to outperform

conv. coding (however block of 40 bits are also possible)

36

Power control

• Important aspect, especially in the uplink

• Near-far problem:

– Codes are not orthogonal or the orthogonality is destroyed by mul-

tipath propagation

– With equal transmit power a MS close to the BS may hide a MS

at the cell border (e. g. with additional 70 dB attenuation)

– Power control has as an objective to control the transmit powers

of the different MS so that their signals reach the BS with the

same level

37

Power control

38

Power control

• Open loop power control: estimate the path loss from the signal

received in DL

• Not accurate: in FDD, UL and DL frequencies are different and fast

fading is uncorrelated between UL and DL

• However used in WCDMA to provide a course initial power setting

• Solution: Fast closed-loop power control: BS performs frequent esti-

mations of the received SIR (Signal-to-Interference Ratio) and com-

pares to a target SIR

39

Power control

• It commands the mobile station to lower or increase its power (in

which cas the mobile causes increased interference to other cells !)

• The command-react cycle is 1500 times per second for each mobile

station (faster than any fading mechanism)

• Also used in DL (no near-far problem however); all signals originate

from the same BS

• Desirable to provide additional power to mobiles closed to the cell

edge

40

Power control illustration

• Uplink, fading channel at low speed

41

Power control in WCDMA

• Fast power control: 1 command per slot or rate 1500Hz

• Basic step size 1dB (also multiples or smaller step sizes)

42

Soft handover

• A mobile is in the overlapping coverage area of two sectors belonging

to different base-stations

• Signals are received from the two BS and recombined by a rake re-

ceiver

• Avoids near-far similar situation when a mobile enters a new cell and

has not yet been power controlled

• Softer handover: same situation but between two sectors belonging

to the same BS

43

Handover

• Previous slide: Intra-mode handover

• Inter-mode handover also supported: the dual mode FDD-TDD ter-

minals may ”handover” from FDD to TDD (measurements mecha-

nisms are implemented)

• Inter-system handover also supported: handover to GSM is currently

only foreseen

44

Transmit diversity

• Closed loop transmit diversity:

– the BS uses two transmit antennas

– mode 1: the terminal feedback controls the phase adjustments (of

the second antenna wrt to the first one) to maximize the power

received at the terminal (FBI field in the slot)

– mode 2: the amplitude is adjusted on top of the phase

• Open loop transmit diversity: space time block coding based transmit

diversity-STTD

45

Block diagram of STTD

46

Transmit diversity

• Assume each transmit antenna is affected by frequency flat fading hi

and fading is constant over 2 signalling intervals

• At time 0 transmit s0 from antenna 0 and s1 from antenna 1

• The received signal for that signalling period is r0 = h0s0 + h1s1 + n0

• At time 1 transmit −s∗1 from antenna 0 and s∗0 from antenna 1

• The received signal for that signalling period is r1 = −h0s∗1+h1s∗0+n1

47

Transmit diversity

• Compute the following combinations (it is not MRC):

t0 = h∗0r0 + h1r∗1 (1)

t1 = h∗1r0 − h0r∗1 (2)

• It comes

t0 = (|h0|2 + |h1|2)s0 + h∗0n0 + h1n∗

1 (3)

t1 = (|h0|2 + |h1|2)s1 + h∗1n0 − h0n∗

1 (4)

• Equivalent to what would be received with one transmit antenna, tworeceive antennas and MRC; therefore diversity order 2

• 2 transmit and M receive antennas would lead to diversity order 2M ,equivalent to 2M MRC (the actual gain depend on many parameters)

48

UTRA

• Complex and efficient technology;

• Still to come ....

49